qwen3.8-max — 價格、上下文與效能

qwen3.8-max 是 Alibaba 的模型，上下文 1M，支援 Reasoning, Tools, Files, Vision。輸入 $2.00，輸出 $6.00，皆為每百萬 tokens。

總覽

2.4-trillion-parameter MoE flagship for coding, professional work, multimodal understanding, and long-horizon agentic workflows

價格

僅列出該模型已開放的套餐。

效能實測

常見問題

模型比較

錯誤排查

429 是限流回應。降低並發、加上退避重試，或切換到配額更高的方案即可。

常見原因

• 並發請求過多
• 目前方案配額已用完
• 失敗後立即重試沒有退避

解決辦法

• 限制並發並以指數退避重試
• 切換到配額更高的方案
• 對重複請求做快取

如何接入

curl https://api.airai.cc/v1/chat/completions \
  -H "Authorization: Bearer <YOUR_API_KEY>" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-max",
    "messages": [{"role": "user", "content": "Hello"}]
  }'

from openai import OpenAI

client = OpenAI(
    api_key="<YOUR_API_KEY>",
    base_url="https://api.airai.cc/v1"
)

response = client.chat.completions.create(
    model="qwen3.8-max",
    messages=[{"role": "user", "content": "Hello"}]
)
print(response.choices[0].message.content)

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: "<YOUR_API_KEY>",
  baseURL: "https://api.airai.cc/v1"
});

const response = await client.chat.completions.create({
  model: "qwen3.8-max",
  messages: [{ role: "user", content: "Hello" }]
});
console.log(response.choices[0].message.content);

const OpenAI = require("openai");

const client = new OpenAI({
  apiKey: "<YOUR_API_KEY>",
  baseURL: "https://api.airai.cc/v1"
});

client.chat.completions.create({
  model: "qwen3.8-max",
  messages: [{ role: "user", content: "Hello" }]
}).then((r) => console.log(r.choices[0].message.content));

將 <YOUR_API_KEY> 替換為令牌設定中的 API Key。

相容端點

Gemini SDK · TypeScript

import { GoogleGenerativeAI } from "@google/generative-ai";

const genAI = new GoogleGenerativeAI("<YOUR_API_KEY>");
const model = genAI.getGenerativeModel({ model: "qwen3.8-max" });

const result = await model.generateContent(
  "Explain quantum entanglement in one paragraph."
);
console.log(result.response.text());

身分驗證

所有請求必須攜帶 Authorization: Bearer <TOKEN> 請求頭；Anthropic 格式端點也接受 x-api-key 請求頭。在「令牌」頁面產生 API Key，可依模型、分組、IP、速率等維度精細化授權。

支援的參數

速率限制

資料更新於: 2026-10-10 18:35